Corporate Failure Prediction (Bankruptcy) in Australia: From Zeta to Neural Networks
56 Pages Posted: 21 Feb 2009 Last revised: 24 Mar 2009
Date Written: February 21, 2009
Abstract
Relatively little has been written on corporate failure prediction in Australia as suitable corporate databases have not been available for research to be undertaken. The authors were given access to suitable commercial databases and have experimented with a number of different techniques of analysis including decision tree analysis, linear regression, discriminant analysis and artificial neural networks. They have then built a predictive model using a neural network approach and this has proved to be strongly predictive. Further work is being undertaken with fuzzy neural networks. Anecdotal evidence indicates that market approaches such as KMVTM and CreditMetricsTM work for the top 50 or 70 stocks in Australia but not generally. We have therefore left such analytics for further experiments. The authors welcome comment about alternative approaches that may be suitable.
Keywords: Bankruptcy diagnosis, Discriminant analysis, Neural networks, Corporate distress risk, Financial ratio analysis, Fuzzy logic
JEL Classification: G33, C49, C88
Suggested Citation: Suggested Citation